Dear All,
It’s an interesting discussion on what might constiute the criteria for a “non-safety related oprp” but the post is in the FSSC22000 safety arena. And perhaps Malaysia has local regulatory standards regarding metal contamination in foods which would then prioritize as far as “acceptable level” is concerned.
As currently stated, the texts of ISO22000 / ISO22004, permit a vast range of interpretations regarding –
(a) Criteria regarding whether a control measure should/could be classified within the CCP/OPRP domain.
(b) if the result of (a) is yes, the criteria regarding whether the CM should be associated with a CCP or OPRP.
© Crteria for validation/verification of the control measure’s ability to acceptably control the hazard.
AFAIK, auditors require to see a documented interpretation of the standard enabling the presentation of criteria for (a,b,c) which seem “logical”, ie can be “validated” (somehow/somewhere/somewhen).
At one time, some posters here noted that their auditors were unhappy at “simple” methods largely based on use of the traditional Codex Decision Tree but this aversion seems no longer evident.
One might consider that the chosen method for finding OPRPs depends on whether few or many CCPs/OPRPs in total are preferred by the user, a route which has some historical haccp echoes.
ISO itself sits nicely on the fence in ISO22004 which indicates that the ratio of CCPs to OPRPs is effectively irrelevant.
George Howlett’s (Safefood) White paper / blog / webinar nicely illustrates/compares some of the OPRP options involved in a user-friendly way although not exclusively oriented to ISO22000. From memory, the White paper includes a discussion on metal detectors as CCPs/OPRPs.
Personally, it seemed to me that the OP had already concluded that the implementation of the 2nd MD was unrelated to a significant food hazard. Therefore, in my ISO interpretation, neither a CCP nor an OPRP. And equally for a 3rd MD if added. 
One could opine that the addition of a final X-ray unit would validate the use of the first MD on its own (assuming X-ray has equal/better sensitivity to MD ?). Or enable removal of both MDs?. And satisfy the auditor ?
Rgds / Charles.C